headers
Richard Earnshaw | 9 Jan 2004 11:04
Favicon

Getting a kernel panic dump back (or not)

Hmm, so I've just hand my shark panic again with the same panic as point 2 
in pr port-arm/23581 :-(

This time, however, on typing sync at the ddb prompt it paniced again (not 
too unusual).  However, on typing reboot it then managed to do a full dump 
(or at least, it went through the motions of doing so).

So I was somewhat disappointed to find that on reboot, savecore announced 
that there was nothing to be found :-(

Anyway, I've DD'ed my entire swap partition (there may be something 
there).  Does anyone have any ideas on how to extract the kernel dump from 
it?  Better still, does anyone know how to make savecore do its job 
properly?

R.
Permalink | Reply | 3 comments |
headers
Robert Swindells | 9 Jan 2004 12:49
Picon
Picon

Re: Getting a kernel panic dump back (or not)


Richard Earnshaw wrote:
>Hmm, so I've just hand my shark panic again with the same panic as point 2 
>in pr port-arm/23581 :-(

>This time, however, on typing sync at the ddb prompt it paniced again (not 
>too unusual).  However, on typing reboot it then managed to do a full dump 
>(or at least, it went through the motions of doing so).

>So I was somewhat disappointed to find that on reboot, savecore announced 
>that there was nothing to be found :-(

>Anyway, I've DD'ed my entire swap partition (there may be something 
>there).  Does anyone have any ideas on how to extract the kernel dump from 
>it?  Better still, does anyone know how to make savecore do its job 
>properly?

The code to dump the core in arm/vm_machdep_arm.c looks identical to the
i386 version. I have used savecore fairly recently on i386.

Can you force it to save using -f ?

Also, savecore seems to want to match the core against the kernel used
to generate it. Perhaps you need to point it at an ELF kernel instead
of the a.out one.

Robert Swindells
Permalink | Reply | 2 comments |
headers
Richard Earnshaw | 9 Jan 2004 12:57
Favicon

Re: Getting a kernel panic dump back (or not)

> 
> Can you force it to save using -f ?
> 
> Also, savecore seems to want to match the core against the kernel used
> to generate it. Perhaps you need to point it at an ELF kernel instead
> of the a.out one.
> 
> Robert Swindells

# /sbin/savecore -f -v -N /netbsd /work
dumplo = 201777152 (394096 * 512)
savecore: no core dump
Permalink | Reply | 0 comments |
headers
Richard Earnshaw | 9 Jan 2004 16:46
Favicon

Re: Getting a kernel panic dump back (or not)

> 
> Richard Earnshaw wrote:
> >Hmm, so I've just hand my shark panic again with the same panic as point 2 
> >in pr port-arm/23581 :-(
> 
> >This time, however, on typing sync at the ddb prompt it paniced again (not 
> >too unusual).  However, on typing reboot it then managed to do a full dump 
> >(or at least, it went through the motions of doing so).
> 
> >So I was somewhat disappointed to find that on reboot, savecore announced 
> >that there was nothing to be found :-(
> 
> >Anyway, I've DD'ed my entire swap partition (there may be something 
> >there).  Does anyone have any ideas on how to extract the kernel dump from 
> >it?  Better still, does anyone know how to make savecore do its job 
> >properly?
> 
> The code to dump the core in arm/vm_machdep_arm.c looks identical to the
> i386 version. I have used savecore fairly recently on i386.
> 

In fact, I think I've found the problem (or one of them).  At no time does 
the ARM dump code set KCORE_MAGIC in the dump, so savecore will never 
recognize it :-(

There's other problems, such as regularly locking up while trying to dump 
as well, but even if it succeeds, it's never going to be recognized as 
things stand.

R.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Franck Baudin | 14 Jan 2004 18:14

Intel Access Library for IXP425

Hi port-arm,

I've looked for Intel Access Library for IXP425 in src tree, and I found 
nothing. This library is mandatory to access ethernet devices integrated 
in IXP425.

So, I've got 3 questions :
- did I miss something ?
- did anyone already use ethernet on IXP425 boards (NOVATEC NTNP425B, 
Intel IXDP425, ADI Coyote) with NetBSD ?
- It seems that Wasaby already ported Intel Access Library on NetBSD but 
did not commit the code 
(http://www.linuxdevices.com/news/NS5787000574.html ) ... Am I wrong ?

Regards,
Franck.
Permalink | Reply | 3 comments |
headers
Robert Swindells | 14 Jan 2004 18:55
Picon
Picon

Re: Intel Access Library for IXP425


Franck Baudin wrote:
>I've looked for Intel Access Library for IXP425 in src tree, and I found 
>nothing. This library is mandatory to access ethernet devices integrated 
>in IXP425.

What does your Intel sales engineer say ?

Robert Swindells
Permalink | Reply | 0 comments |
headers
Jason Thorpe | 14 Jan 2004 22:47

Re: Intel Access Library for IXP425


On Jan 14, 2004, at 9:14 AM, Franck Baudin wrote:

> I've looked for Intel Access Library for IXP425 in src tree, and I 
> found nothing. This library is mandatory to access ethernet devices 
> integrated in IXP425.

Correct.  The Intel Access Library for IXP425 is not in the source tree.

> So, I've got 3 questions :
> - did I miss something ?

Nope, you're exactly right.

> - did anyone already use ethernet on IXP425 boards (NOVATEC NTNP425B, 
> Intel IXDP425, ADI Coyote) with NetBSD ?

Wasabi has NetBSD ports to several IXP425-based boards that support the 
IXP4xx Ethernet and the IXP4xx crypto processor.

> - It seems that Wasaby already ported Intel Access Library on NetBSD 
> but did not commit the code 
> (http://www.linuxdevices.com/news/NS5787000574.html ) ... Am I wrong ?

The Intel Access Library for IXP425 requires a license from Intel, and 
thus cannot be contributed to the public NetBSD sources.

         -- Jason R. Thorpe <thorpej <at> wasabisystems.com>
(Continue reading)

Permalink | Reply | 1 comment |
headers
Franck Baudin | 15 Jan 2004 09:39

Re: Intel Access Library for IXP425

Jason Thorpe wrote:

> The Intel Access Library for IXP425 requires a license from Intel, and 
> thus cannot be contributed to the public NetBSD sources.

But perhaps it is possible to manage it as a vendor contribution : the 
developper has to obtain an Intel license (fortunateny I've got one ;-)  
), but some patches and makefiles could be available in the NetBSD 
community, like snapgear-3.0.0-ixp400-v1.3-support.tar.gz for Linux that 
you can find there: 
http://ixp4xx-osdg.sourceforge.net/howtos/snapgear-3.0.0-IXP400-AccessLibrary-v1.3-Integration-HOWTO.html

Is there any chance to find this as open source, or do I have to write 
it from scratch ?

Regards,
Franck.
Permalink | Reply | 0 comments |
headers
Port-arm32 | 15 Jan 2004 21:12

Everything at the touch of your fingertips! fittingly workhorse

aflame screws
shampoo involve

Now you can order your medications online and 
have them delivered to your doorstep overnight.

Why leave your house? All the known m.e.d.s 
you need at the touch of your figertips at the 
lowest cost you will ever see.

recesses randomize
tantrums humidly
Buchanan dactylic
tinniness chords

Link yourself below to see for yourself!

imprints http://www.webrxcentral.com household
Permalink | Reply | 0 comments |
headers
Richard Earnshaw | 17 Jan 2004 17:03
Favicon

Current ARM kernels hang during bootstrap

Jaromir,

Since you committed your exit-path patch on 2004/01/04 all my ARM systems 
have started to hang during bootstrap.

Investigation using a Multi-ICE box shows that the vectors page is getting 
unmapped (on all my machines the vectors are at VA 0x0, which puts it 
notionally in user memory, although it's really a kernel page).  Loosing 
the interrupt and fault vectors is obviously somewhat fatal to the health 
of the machine.

Have you any idea what might be the cause of this change in behaviour?

R.
Permalink | Reply | 5 comments |

Gmane